By Sloane, PR and Marketing, HUMAN HIVE
Are your people using AI at work without any rules from you? Very likely: a survey released November 11, 2026 found 48% of Canadian office workers have never received guidance from their employer on how to use AI safely, and 39% use free or personal AI accounts for work. The fix is not a ban. It's a one-page set of rules, a short hands-on session, and leaders who go first.
The 60-second version
Nearly half of office workers surveyed say their employer never told them how to use AI safely at work.
When no one sets the rules, people pick their own tools, and company data goes wherever the free tool takes it.
This week, write a one-page AI guide with three lists: tools you can use, data you never paste, and who to ask.
What happened
On November 11, 2026, SecurityInfoWatch and Canadian Manufacturing reported a new survey commissioned by ISA Cybersecurity. Angus Reid Forum surveyed 1,042 Canadian office workers in October about cybercrime and how they use AI at work.
The headline: 48% have never received guidance from their employer on how to use AI safely at work. Meanwhile, use keeps climbing. 71% say they moderately or significantly increased their AI use at work over the past year. 47% use tools like ChatGPT, Claude, Copilot or Gemini for work at least a few times a week, including 24% who use them daily. And 39% use free or personal AI accounts for work.
Policy is just as fuzzy. Half say their employer has a clear AI policy. 18% say the policy is unclear, 17% say there is none, and 15% aren't sure.
Andrew Buckles, executive vice president of services at ISA Cybersecurity, summed it up: "Canadian workers have adopted AI faster than their employers can enable and train them to use it safely." He also flagged a risk inside approved tools: "If an employee can reach more data than their role requires, so can the AI assistant working on their behalf."
This isn't only a Canadian story. Microsoft and LinkedIn's 2024 Work Trend Index, a survey of 31,000 people across 31 countries, found "78% of AI users are bringing their own tools to work" and "only 39% of users have received AI training from their company." Different survey, different year, same pattern: people adopt first and get trained later, if at all.
Why it happened (the human side)
Nobody in these numbers is a villain. People found a tool that saves them an hour, and nobody told them where the lines are. So they drew their own.
Three human habits keep the gap open:
Leaders wait for the perfect policy. Legal and IT want something airtight, so nothing ships, and the team keeps improvising.
Silence reads as permission, or as danger. Some people assume "no rule" means anything goes. Others assume it means "don't get caught," so they hide their use.
Training gets treated like an event. A webinar in March doesn't help someone deciding at 4 PM on a Thursday whether a client contract can go into a chatbot.
What Mitch has taught on this
Mitch has been telling rooms about this for a while. In his CMCA 2025 keynote, answering the question "Can my competitors see what is put in AI?", he warned about free tools: "if you're using a free version, it's the old saying in social media, if it's free, it's about me." Settings and terms differ by tool and plan, so check each one, but his rule of thumb holds: know what a tool does with what you give it before you give it anything.
He was just as clear that fear isn't the answer. On AI detectors and policing people: "Don't try to detect others. Try to open up an environment where people feel safe using a tool in the right way." And on choosing tools: "You only want to use models that are proven and trusted."
That's the whole playbook in three lines. Name the trusted tools. Make it safe to ask. Teach the right way, out loud.
Best practices
Ship a one-page AI guide this week, not a perfect policy next quarter. Owner: the department head, with one person from IT or security. Three lists: approved tools (and which account or plan), data that never goes into any AI tool (client contracts, personal data, passwords, financials), and the name of the person to ask. Version 1 can be wrong in places. Update it monthly.
Give people a paid, approved option. If the only tool that works is a personal account, people will use their personal account. Pick one approved tool per team and show people how to sign in.
Run a 30-minute "real work" session. L&D or a team lead brings three real tasks from this week: an email, a summary, a spreadsheet. Do them live with the approved tool, and point out the moment where data rules apply.
Trim AI permissions to the job. Buckles' warning applies to approved assistants too. Ask IT one question: can our AI assistant reach files this person doesn't need? If yes, fix access before you expand the tool.
Make "I'm not sure" a normal question. Set up one channel or inbox for AI questions and have a leader answer the first few publicly. No blame for asking after the fact.
What's driving people crazy this week
Rules that show up after the habit. Workers have been using AI for a year, and many are only now hearing they may have been doing it wrong. The ISA survey puts numbers on it: use up for 71%, guidance missing for 48%. It echoes a Jobs for the Future survey we logged this month, where 56% of workers said their employer never asked how AI tools are used in their work. The human fix: when you roll out the rules, start with "thank you for figuring this out on your own. Here's how we make it safe," not with a warning.
What this means for the next 1 to 3 years
Our bet: the guidance gap closes from the bottom up, not the top down. The teams that get this right will treat AI rules like a recipe card taped to the fridge, short, specific and edited often, not like a binder on a shelf.
Three things we expect by 2028. First, "AI onboarding" becomes part of every new hire's first week, right next to email and the badge, because the tools will be part of the job on day one. Second, the most trusted people on a team will be the ones who can explain why they didn't put something into AI. Discretion becomes a visible skill. Third, AI assistants will get scoped like employees: a role, a set of files, a manager. The human who sets those limits holds one of the most important jobs in the building. That's the synergy we care about: AI moves fast, and people decide where it's allowed to go.
How we evolve
The human skill to build next is discernment: the quick, everyday judgment of what to share, with whom, and why. It matters more as AI gets better, because a better tool makes it easier to hand over more, faster, without thinking. Mitch's philosophy is Amplify, Not Just Automate. You can't amplify good judgment you never practiced.
One small daily practice: before you paste anything into an AI tool today, pause for three seconds and ask, "Would I be fine if this showed up in a stranger's inbox?" If the answer is no, strip the names and numbers first, or don't paste it. Do that for two weeks and it becomes a reflex.
Try this today
Ask at your next team meeting (5 minutes): "Which AI tools are you actually using for work, and on which account?" No judgment. Write down the list.
Draft the never-paste list (10 minutes): five kinds of information that never go into any AI tool. Share it the same day.
Send one note (10 minutes): use the script below to set the tone before any policy arrives.
Free tool: the "AI rules, version 1" note
"Quick note on AI at work. Thank you to everyone who's been figuring this out on your own. Here's version 1 of our rules. Use: [approved tool and account]. Never paste: [client names and contracts, personal data, passwords, financials, anything marked confidential]. Not sure? Ask [name] in [channel], no blame, even after the fact. We'll update this every month, and I want your edits."
FAQ
What percentage of workers have no guidance on using AI safely at work?
In a survey of 1,042 Canadian office workers conducted in October 2026 by Angus Reid Forum for ISA Cybersecurity and reported on November 11, 2026, 48% said they had never received guidance from their employer on how to use AI safely at work.
What is shadow AI?
Shadow AI is the use of AI tools at work that the employer hasn't approved or can't see, such as personal or free accounts. In the same survey, 39% of respondents said they use free or personal AI accounts for work.
Should companies ban personal AI tools?
A ban without a good approved option tends to push use out of sight. A better first step is a one-page guide: approved tools, data that never goes into AI, and a named person to ask, plus a short hands-on session on real work.
Where HIVE can help
Planning 2027 learning? Talk with us about training your people will actually use on Monday. Not sure where your team stands? Take the free readiness assessment.